Podcast episode

Can Content Drive Performance?

by Ken Rona ·July 10, 2026 ·AdExchanger: The Big Story

brand-safety content-quality ctv dsp programmatic

TL;DR

AdExchanger's editorial team covers two converging trends: the rise of AI-generated "slop" content as a successor threat to made-for-advertising (MFA) inventory, and a buyer-built technical workaround that unlocks show-level transparency on connected TV (CTV) buys. Both stories argue that content quality — not just audience targeting — is becoming a measurable driver of campaign performance.

What was covered

AI slop and MFA trends: The ANA (Association of National Advertisers) quarterly ad-spend benchmark reports show MFA (made-for-advertising — low-quality sites built to maximize ad load) spend fell from 15–21% of budgets in 2023 to roughly 0.4–0.6%, then ticked back up to ~1.1% in the most recent quarter. The uptick is attributed partly to AI-generated low-quality content flooding supply.
AI slop ≠ MFA (technically): Industry experts draw a distinction: formal MFA requires low-quality content plus high ad load plus paid traffic arbitrage. AI slop typically lacks the paid-traffic component, so it doesn't always qualify under the industry's definition — but SSPs say the distinction doesn't matter operationally; it's all low quality and has to go.
Social media as an unaddressed slop vector: Hosts note that AI slop on Facebook, Instagram, and TikTok — fake viral videos, fabricated text exchanges — is largely absent from ANA's analysis, which focuses on open-web MFA sites. YouTube, Meta, and X have taken some steps, but coverage is incomplete.
AI makes MFA site creation faster but potentially less durable: Sites spin up and are abandoned so quickly that operators may not have time to build the ad infrastructure that older MFA sites had — a modest structural silver lining.
CTV show-level transparency workaround: Peer 39 (a contextual advertising platform) has ID integrations with CTV publishers that generate anonymized content IDs tied to individual shows. These IDs pass through the SSP without exposing viewer identity (addressing VPPA — Video Privacy Protection Act — concerns), then get ingested by Pontiac Intelligence's DSP (demand-side platform, the software advertisers use to buy digital ads). The agency Intermedia Advertising used the setup to calculate cost-per-acquisition by show, enabling mid-flight optimization.
Publisher opacity and mislabeling: Mario Diaz, CEO of Peer 39, reported that during Q4, inventory labeled as "documentary programming" by a publisher was actually the King Charles/Queen Elizabeth eulogies. Advertisers unknowingly bid CPMs comparable to live sports for that inventory.
Buy-side trend toward pre-bid content signals: The Trade Desk is integrating attention metrics and media quality signals into pre-bid decisioning; Amazon DSP is doing the same. Viant, iSpot.tv, and Beeswax are developing similar show-level content data integrations for buy-side decisioning.

Notable claims & predictions

Victoria McNally (AdExchanger): "AI slop can be MFA, but it's not always automatically MFA — the proper definition of MFA requires low-quality content, really high ad load, and paid traffic. You need all three."
Anthony Vargas (AdExchanger): David Nurenberg at Intermedia Advertising said the best-case scenario when a publisher provides show-level data is "maybe 10% to 15% of the impressions for a campaign are going to have some kind of show-level data or even genre-level data attached to it" — meaning 85%+ of CTV impressions remain opaque even in favorable scenarios.
Anthony Vargas (AdExchanger): Publishers sometimes mislabel inventory intentionally to protect pricing — the eulogies-as-documentaries example illustrates that self-reported content classifications can be unreliable enough to drive CPMs to live-sports-equivalent levels for irrelevant content.
Sarah Sluis (AdExchanger): "Buying through a DSP has always been viewed as a performance medium…buying by audience instead of content. But now we're seeing this pendulum swing back where media quality — whether it's context, content, attention, or not being MFA — is also being taken into consideration."
Victoria McNally (AdExchanger): Scott Pearson (now at The Trade Desk, previously at IAS) said a Last Week Tonight with John Oliver segment on AI slop on social media was a turning point that caused clients to start actively asking about AI slop in their buys.

Fact check

Claim (Victoria McNally): MFA's share of ad spend started "between 15 and 21 percent" in early ANA reports, then fell to ~1.1%, having previously hovered between 0.4–0.6%. Assessment: Unverified in precise figures. The ANA/Ad Fraud Alliance's 2023 MFA research is real and widely cited, but the specific percentages quoted here — and the trajectory — could not be confirmed against the original reports from this transcript alone. The 15–21% figure is a plausible range cited in industry press at the time, but listeners should verify against the actual ANA benchmark report rather than treating these numbers as precise.
Claim (Anthony Vargas): The VPPA ("blockbuster law") is the key legal barrier preventing publishers from sharing show-level viewing data with ad buyers. Assessment: True but omits context. VPPA does restrict disclosure of personally identifiable video viewing records and has been actively litigated in the streaming context (multiple class actions against streaming services). However, hosts also acknowledge — and Vargas implicitly agrees — that channel conflict and commercial leverage are at least as significant as legal concern in publishers' reluctance to share show-level data. Framing VPPA as the primary barrier flattens a more complex picture.
Claim (Sarah Sluis): The difference in MFA exposure between best- and worst-performing marketers is "between maybe 1% MFA and 2% MFA" — best performers under 1%, worst performers over 2%. Assessment: Unverified; talking their own book. These figures are attributed to the ANA benchmark report, but the specific performance-correlation framing (best vs. worst marketers segmented by MFA %) was not sourced to a direct quote or published finding in the transcript. The ANA has a clear institutional interest in highlighting MFA's harms to justify continued measurement investment. Listeners should pull the underlying report to assess the methodology.
Claim (Anthony Vargas / Mario Diaz): Peer 39 detected in Q4 that inventory labeled "documentary programming" was actually eulogies (for British royals, implied), with CPMs bid up to live-sports levels. Assessment: Unverifiable but plausible. This is a specific, named-source claim from Peer 39's CEO — a company that profits from selling content classification transparency. The anecdote is directionally credible (publisher mislabeling is documented), but Peer 39 has an obvious commercial incentive to highlight dramatic examples of the problem their product solves. No independent corroboration in the transcript.

Why this matters for ad-tech operators

For DSPs and buy-side platforms: The show-level transparency workaround described — Peer 39 content IDs passed through SSP pipes into a DSP for mid-flight decisioning — signals a near-term competitive axis. The Trade Desk, Amazon DSP, Viant, and Beeswax are already moving on pre-bid content/attention signal integration. DSPs that can't ingest and action these signals will be at a disadvantage as agencies demand mid-flight CTV optimization, not just post-campaign reporting.
For SSPs and CTV publishers: The eulogies

Full analysis

Two things are happening at once in this episode, and they rhyme. First, the industry's old enemy — made-for-advertising junk sites — has been beaten down to near zero, only to be replaced by AI-generated "slop" that the current rulebook doesn't even classify as a problem. Second, a scrappy technical workaround is finally cracking open the black box of connected-TV buying, letting advertisers see which show their ad ran against. The connective tissue: after fifteen years of buying audiences and ignoring content, the industry is swinging back toward caring about where the ad actually lands.

The decision this poses for operators is not "should we care about content quality" — that ship has sailed. It's how much to invest, how fast, in content-level signals when the data coverage is thin and the definitions are shifting under everyone's feet. This is a Type 2, reversible bet for most players: you can start ingesting content signals and dial up or down. The forcing function is competitive, not regulatory — The Trade Desk and Amazon DSP are already moving.

The Market Analyst. Watch who's converging here. The Trade Desk and Amazon DSP baking attention and media-quality signals into pre-bid decisioning is the tell — the two biggest buy-side platforms are declaring content quality a first-class input, not a post-campaign report card. That pulls the whole verification layer (DoubleVerify, IAS) upstream and squeezes anyone selling content quality as a standalone after-the-fact audit. Meanwhile Peer 39, Viant, iSpot, and Beeswax are racing to own the CTV show-level data pipe. For the generalist: the giants are turning "was my ad in a good place" from a receipt you get later into a decision made in the split second before you buy. Whoever owns that signal owns leverage.

The Skeptic. The load-bearing assumption is that content quality actually moves performance — and the evidence here is thin and self-interested. The ANA numbers are unverified. Peer 39's royal-eulogies anecdote comes from the company that sells the fix. And bury this stat from the episode: even in the best case, only 10–15% of CTV impressions carry show-level data. That means 85%+ stays opaque. You cannot optimize a campaign on 12% coverage; you can only cherry-pick anecdotes from it. For the generalist: someone selling you X-ray glasses just showed you one dramatic X-ray, and the glasses only work on one wall in ten.

The Operator. The eulogies-labeled-as-documentary problem is the real story, and it doesn't get solved by a clever ID pass-through. It gets solved when publishers stop mislabeling — and they mislabel on purpose to protect pricing. The episode's own fact-check flags that channel conflict, not the VPPA privacy law, is the bigger barrier. So the Tuesday-morning reality: a buyer wires up Peer 39 IDs into a Pontiac DSP, gets clean data on a sliver of inventory, and still eats blended CPMs on the opaque majority. Second-order effect at 90 days: publishers who do share data get punished with lower bids on their weak shows, which teaches everyone else to keep the lights off.

The Customer / End User. Advertisers didn't wake up demanding this — a John Oliver segment scared them into it. That matters. Client interest in AI slop is reputational panic, not performance rigor. The CMO who now asks "is my ad next to fake AI videos" mostly wants a clean answer for the boardroom, not a 3% CPA improvement. For the generalist: the buyer's real fear is a screenshot of their brand next to garbage going viral, not a rounding error in cost-per-sale. That's a brand-safety sale dressed up as a performance sale — and it prices very differently.

The CFO. Two separate P&L stories here. Slop cleanup is cheap and defensive — SSPs say it's all low quality and has to go, so filtering it is table stakes with near-zero upside. Show-level CTV transparency is the expensive, offensive bet: integration work, DSP ingestion, agency services layered on top, all to illuminate 12% of impressions. The payback only pencils out if that 12% is where the money concentrates (premium sports, tentpoles) and if better placement genuinely lifts outcomes enough to justify paying up. Otherwise you've built a Ferrari to drive one block.

Where the council splits:

Is content quality a performance driver or a brand-safety hedge? The Market Analyst says the platform convergence proves it's becoming a real buying input. The Customer says the demand is reputational fear triggered by a comedy segment. These price completely differently — one is a durable optimization market, the other is a defensive line item that fades when the news cycle moves on.
Does the CTV transparency workaround scale, or stay a boutique trick? The Market Analyst sees an emerging competitive axis; the Operator and CFO see 12% coverage and publishers structurally motivated to keep it low. If publishers won't willingly label honestly, technical cleverness only illuminates the inventory nobody was worried about anyway.
Is AI slop a new threat or the old MFA fight relabeled? The episode notes slop often lacks the paid-traffic arbitrage that defines true MFA — and that AI sites spin up and die too fast to build ad infrastructure. That "structural silver lining" suggests the threat may be noisier than it is durable.

What it actually hinges on: whether publishers will surrender show-level content data at scale. Everything downstream — the DSP signal race, the transparency workarounds, the CPA-by-show optimization — is gated by supply-side willingness, and the supply side has a direct financial incentive to stay dark. The technology (anonymized content IDs passing through SSP pipes, sidestepping the VPPA video-privacy law) is solved. The commercial standoff is not.

The council leans skeptical on scale, bullish on direction. Content-level buying is genuinely where the puck is going — the two dominant DSPs moving in unison isn't noise. But the coverage math means this is a multi-year grind, not a 2026 unlock. Operators should treat slop-filtering as cheap hygiene to do now, and treat show-level CTV transparency as a real-options bet: build the ingestion capability, run pilots on premium inventory where the data actually exists, and don't over-promise mid-flight optimization to clients when 85% of the buy is still a black box.

Before committing budget: verify the ANA figures against the actual report (the episode's own fact-check flags them as unconfirmed and the ANA benefits from the scary numbers), and pressure-test any vendor's coverage claim against your real CTV mix, not their demo reel.

Prediction: By the IAB or ANA's next major benchmark report in early 2027, no leading vendor or industry body will report show-level content data coverage exceeding 25% of CTV impressions — the publisher-mislabeling and channel-conflict problem will keep coverage stuck in the low double digits.

Confidence: Medium — the episode's own best-case number is 10–15%, and publishers are structurally motivated to stay opaque.

Why: Show-level transparency requires publishers to volunteer data that lowers their pricing leverage; the eulogies-as-documentary example shows they actively obscure it. A technical workaround on the buy side can't force honest labeling on the sell side, so coverage stays thin regardless of DSP demand.

Revisit by 2027-03-31: We're right if the next ANA/IAB CTV transparency benchmark (or a Peer 39/Viant/iSpot public claim) still shows show-level data on under 25% of impressions. We're wrong if any credible source reports coverage above 25%, signaling publishers are opening up faster than their incentives suggest.

Listen to the episode →